Hi Guys, I have 2 1660 supers with Hynix memory that were running hot at around 63/64 and high power of 86 and 88 watts using -250 and -1004 memory setting. I matched the PL for them (This is a 5 card rig 4 1660 supers and one 1660 regular) The 1660 regular is at -200 and +1300 and is fine…
I changed the two Super cards to +930 for the Core and left -1004 and now the watts on the cards are 73 and 75. Temps are now 45 and 55 degrees which is a significant drop.
Now question is this, changing the core from -250 to plus 930 does what in term of stressing the card limits? Ok temps and watts have lowered significantly thats for sure.
Is it better to keep at -250 and have higher power and temps as maybe the GPU core is now running beyond spec etc?
Im curious as to why these 2 cards (Asus Tuf Duals) dont like the usual settings like all my other 1660 supers with Hynix as I have aother rig with 4 of them… MSIs and Another Asus Tuf and their watts are 80 and temps in the mid 50s… just these two cards want more power and run way hotter.
I do have these two card risers connected to a PCIE extension/adapter card in the main X16 Slot so not sure if this is causing it?
I usually see 1660 supers with Hynix Memory using -250 core and -1004 memory with PL of 80 which is what i have been using… but changing to +930 core really dropped the watts and temps…
Well i have tried the two settings… pretty much the consensus for the 1660 super is -250 core and -1004 memory PL80ish… which work for all my cards for the most part… just have really one problematic at that setting, and its Hynix memory… 63 degrees, 88 watts its at… i use the 930 for the core then bam temps go down, PL goes to 73…
ill give it a go… any ideas why a higher core clock speed would lower temps and power to such a degree? i would have thought it would be the oppsite affect by doing that.
If you set an absolute core clock for Nvidia cards then there shouldn’t be any reason to also set a PL. I have a 1660 Super with Hynix memory and it runs at 73W at 31.8mhs with 1000 on core and -1010 on memory.
yeah i put these settings and it lowers the watts and temps big time… but whats the counter affect of having a conttant core clock of 1000+ rather than a negative -250 like most 1660 supers that are set at…
Having a constant core clock is better. You do not use much core when mining ETH but the core still boosts according to its boost logic which takes up unnecessary power and produces unnecessary heat.
If you run the command nvtools --clocks with an offset you will see your core clock is much higher than 1000mhz which is not needed.
Setting a constant core removes fluctuations of core during mining giving you more stability and less power/heat increases the lifespan of your card.
You can try setting your core clock for your non super 1660 to 1300 and give it a go. It should also significantly reduce power/temp.
After… im having troubles with the regular 1660… ive tried a few settings and just need to figure out what to use for the Samsung and Hynix… someone mentioned like 1500 for GOU and 1500 memory for Samsug but that crashes my rig…
I would like to get the one that has the 61degree temp using the Absolute core setting… that way i can also probably lower the fan percentage as the temps would probably rop also.
